基于药物相似性网络数据的药物组合预测方法 提出一种基于图表示学习进行药物组合治疗的预测方法

NEWMIN method is a drug combination prediction algorithm based on graph representation learning. It included four main steps: 1) Constructing the multilayer drug similarity network . 2) Evaluating the importance of each network in the multilayer network . 3) Sampling multiplex networks information . 4) Feature learning. After using NEWMIN in the multiplex network,  the embedded vector representation of the nodes can be obtained, which helped the researchers use machine learning algorithms to make more accurate link predictions, such as drug-drug interactions.